You have decided to qualify. The form is open, the cursor is in the first empty label, and the honest position is that you do not know what to type into it.
Most advice hands you a list of thirty questions to pick from, which is the opposite of useful. The number of questions is the decision. The wording is the craft. Get the count wrong and no phrasing rescues it.
Three questions, and what the fourth one costs
Every question is a toll, the person paying it has not met you yet, and each is charged on top of the ones in front of it. What is less obvious is that the toll is not collected evenly.
Take the plain part with arithmetic rather than a statistic. Pick a number for how many people give up at each extra question. Say one in ten, invented, so substitute your own and watch the shape hold: three questions leaves roughly 73 of every 100 who started, six leaves roughly 53.
The uneven part is a claim about your market rather than a fact about forms, so check it rather than trusting it. Whoever abandons at question four is usually the person with the least patience for a form, and that correlates with having a diary somebody else manages. Your questions lean towards keeping the people with time on their hands.
The ceiling is low and the answer is boring. Three is usually right. Five is the ceiling. Setupp will let you write thirty, and the validation message says what it thinks of that: thirty questions is far more than anyone will answer.
Ask for a band, not a number
Nobody answers "what is your budget?" honestly into an empty box. There is no honest answer available, because a number typed into a stranger's form is not information, it is an opening position. So they leave it blank, or write the figure they think buys the meeting.
Give them a list instead, with the lowest option below your floor. The question stops being a negotiation and becomes a multiple choice, and picking the bottom band costs nothing, because somebody clearly put it there on purpose.
The second reason is mechanical and it is the bigger one. A band can be scored; free text cannot, at least not reliably. An option list is written a line at a time, with an optional weight on the end.
Under $5,000 = 0
$5,000 to $15,000 = 3
$15,000 to $30,000 = 6
$30,000 and above = 9Those weights are what the scoring engine reads. A pick-one field is also the most rule-friendly shape on a form: "is one of" there means exactly what it looks like. The editor will not even offer contains on a pick-one, because against a list the booker never types into it is either the same rule or one that never fires, and a rule that never fires is worse than none. It looks armed.
The questions people phrase wrongly
One mechanism sits underneath almost every badly worded qualifying question. A bad question asks somebody to admit something. A good question asks them to describe something. Admitting is a status move and people decline to make it in front of strangers. Describing is talking about their own job.
- "Are you the decision maker?" becomes "Who else would be involved in this decision?" The first gets a yes from everybody, including whoever will take it to a committee in March. The second gets you an org chart.
- "How urgent is this?" becomes "What made you start looking now?" Every self-assessment on a booking form comes back high. What triggered the search is a story, and the story either has a date in it or it does not.
- "Are you happy with your current provider?" becomes "What are you using today?" The first asks somebody to criticise a decision they probably made. The second gets the same answer plus the switching cost.
- "Do you have budget approved?" becomes the band. Approval is a yes or no about a process you cannot see. The band tells you the size of the thing, which was the part you needed.
Read each question and ask what the person has to concede in order to answer it truthfully. If the answer is anything at all, rewrite it as a description.
One question that is not qualifying anything
Every form should carry one open question with no filtering job at all. Its only purpose is to give whoever takes the call something to prepare against, and it is usually worth more than a third filter.
It is a long-text field: "what do you need help with?". A fractional CFO reading "Series A readiness" in that box is the whole example. Nothing there decides whether the meeting happens, and it changes everything about the first ten minutes. It turns a booking into a briefing.
The product is honest about how it treats that box. With no AI key configured, the scorer grades free text on length and whether there are figures in it, and the note says as much: answered, but thinly, with little for you to prepare against. A short answer is not a bad lead, it is an unprepared call.
Make it optional. A required paragraph is the most expensive field on any form, and the people who abandon at one are the busy ones again. It still gets answered by whoever wants the call to go well, and three sentences typed into a box they could have skipped means they have spent something on the meeting, which is the mechanism underneath no-shows more than reminders are.
Block, flag, and what the score does with the rest
Every rule has to pick one of two jobs, and this is where first drafts go wrong.
- Block is for a floor you will not move. The calendar never renders, the booker gets the message you wrote, and nothing about them is stored beyond which rule said no and when.
- Flag lets the booking through with the score penalised and the booking marked. The meeting still happens. You arrive at it already knowing.
Flag is right more often than people expect, because most answers are signals rather than disqualifiers. Spend below your floor is a disqualifier. A timeline of nine months, a channel you do not run, a stack that is adjacent rather than matching: each is worth knowing, none is worth turning somebody away over.
A flag costs arithmetic you can check: 1.5 off the score each, at most three counted, so the worst discount is 4.5 out of 10. A discount, not a verdict.
The score under it is a weighted mean. Write three or four criteria in your own words with a weight on each, relative rather than percentages, so 3, 2 and 1 mean the same as 30, 20 and 10. Each criterion is matched to the question that answers it, by key first and word overlap second. Where nothing matches, the note says the value was inferred rather than quietly inventing a number.
The reasoning comes back written rather than numbered. It names the strongest signal and the weakest, and below 7 out of 10 it names the single biggest risk with an instruction attached: qualify hard on that early in the call. None of it is ever shown to the booker.
Four floors, and how to tell yours is wrong
Four setups, four floors, and in none of them is the floor really about the answer. It is about what the wrong meeting costs that particular business.
- A [DTC agency](/for/agency) asks for a monthly ad spend band. Both partners sit on the discovery call, so a brief under the floor costs an hour and a half of the two most expensive calendars in the business. The floor is whether the retainer covers the cost of finding out.
- A [sales team](/for/sales) asks for headcount as a band. The floor is seats, set where the deal stops clearing the cost of the AE who would work it. Below that the answer is not no, it is the self-serve page.
- A [recruiter](/for/recruiting) asks for years of experience and the primary stack. Years is a plain number, not a band, because it is a fact nobody negotiates. The floor is four on a senior requisition and zero on the graduate one, which is why forms attach to the event type rather than the account.
- A [consultant](/for/consulting) asks for an engagement budget band. The floor is the smallest engagement worth the delivery it will take. Nothing stands between the enquiry and the call, so a small project found forty minutes into an afternoon comes out of delivery.
None of them can be copied, so the only way to check yours is to look at what it did. Setupp's declined log holds which rule said no and when. Not the name, not the address, not the answers, not a hash of any of them, which is the honest position for somebody who filled in a form and got a no. It is also sufficient, because what you need is not who was turned away. It is how many, and by which question.
Read it after a fortnight and ask one thing. Is a rule turning away people you would have met? If so, the rule is wrong. Not the approach, not the form, not the idea of qualifying. One threshold, usually one band too high, usually the money one.
The case for putting all this in front of the calendar rather than behind it is made properly in the pillar post. A beautifully worded form that runs after the slot is taken is still a form that runs after the slot is taken.
Setupp puts the questions, the rules and the score in one place rather than three that have to be wired together. The form and the scoring are on the free tier; the gate and the declined log start at Pro, on the pricing page. Whichever tool you use, the shape is the same. Write three questions, make one a band and one open, block on exactly one thing, then read the log in a fortnight.